Carrie Henning-Smith1, Shailendra Prasad2, Michelle Casey1, Katy Kozhimannil1, Ira Moscovice1. 1. Division of Health Policy and Management, Rural Health Research Center, University of Minnesota School of Public Health, Minneapolis, Minnesota. 2. Department of Family Medicine and Community Health, University of Minnesota School of Medicine, Minneapolis, Minnesota.
Abstract
PURPOSE: Quality scores are strongly influenced by sociodemographic characteristics and health behaviors, many of which lie outside of the clinician's control. As a result, there is vigorous debate about whether, and how, to risk-adjust quality measures. Yet, rurality has been largely missing from this debate, even though population and environmental characteristics are demonstrably different by rurality. We addressed this gap by examining the influence of county-level population sociodemographic, environmental, and health characteristics on 3 Medicare quality measures. METHODS: We used a cross-sectional analysis of 2016 County Health Rankings data to estimate differences in 3 Medicare quality scores (preventable hospitalizations, HbA1c monitoring, and mammography screening) by rurality. We then adjusted for county-level sociodemographic and environmental characteristics in multivariable regression models in order to see whether the association between rurality and quality was impacted. FINDINGS: Both micropolitan and noncore counties exhibited lower quality scores than metropolitan counties for all 3 measures. After adjustment, noncore counties still had poorer quality on all 3 measures, while micropolitan counties improved on 2 measures. Several county-level sociodemographic and environmental characteristics were associated with quality, although the direction of association depended on the quality measure. CONCLUSIONS: Differences in Medicare quality scores by rurality cannot be entirely explained by differences in population or environmental characteristics. Still, to the extent that clinicians are evaluated-and paid-based on measures that are influenced by both population sociodemographic characteristics and geographic location without adequate risk adjustment, the challenges of delivering care in rural areas will only be exacerbated.
PURPOSE: Quality scores are strongly influenced by sociodemographic characteristics and health behaviors, many of which lie outside of the clinician's control. As a result, there is vigorous debate about whether, and how, to risk-adjust quality measures. Yet, rurality has been largely missing from this debate, even though population and environmental characteristics are demonstrably different by rurality. We addressed this gap by examining the influence of county-level population sociodemographic, environmental, and health characteristics on 3 Medicare quality measures. METHODS: We used a cross-sectional analysis of 2016 County Health Rankings data to estimate differences in 3 Medicare quality scores (preventable hospitalizations, HbA1c monitoring, and mammography screening) by rurality. We then adjusted for county-level sociodemographic and environmental characteristics in multivariable regression models in order to see whether the association between rurality and quality was impacted. FINDINGS: Both micropolitan and noncore counties exhibited lower quality scores than metropolitan counties for all 3 measures. After adjustment, noncore counties still had poorer quality on all 3 measures, while micropolitan counties improved on 2 measures. Several county-level sociodemographic and environmental characteristics were associated with quality, although the direction of association depended on the quality measure. CONCLUSIONS: Differences in Medicare quality scores by rurality cannot be entirely explained by differences in population or environmental characteristics. Still, to the extent that clinicians are evaluated-and paid-based on measures that are influenced by both population sociodemographic characteristics and geographic location without adequate risk adjustment, the challenges of delivering care in rural areas will only be exacerbated.